Processing ROOT histogram Files¶
The (current) need for Rebinning¶
The HARP and ITEP analyzer modules fill histograms with a fine binning. At the end of the job they also fetch the Dossier data files that they are told about (see here). This (at least at the time) meant that the data and the MC histograms had different binning. This isn't something expected by the
professor fitting structure so had to be dealt with.
I'm told that newer versions of the analyzers will rebin to match the Dossier data ... if they get it. I had systematic trouble getting Dossier histograms for all cases (possibly due to the system being overwhelmed when all analyzers in a job simultaneously request those histograms (since the source feeding Geant4 modules with a primary particle reaches its limit and triggers the final stages of the job, and when other jobs in the cluster, which should have similar run times, also complete). So until there is significantly more robustness of this data fetching this problem is here to stay.
show the file structure of root files out of analyzers ...
later show the concatenated file structure ...
The strategy chosen here is to collect all the histogram files for all universes for a given "exptsetup" and assume at least one of them retrieved all the requisite data histograms -- in practice this turned out not to be true, so some distributions weren't represented in the final data set (this appears to be the case for 26 of the 517 distributions it possibly could have with the (itself flawed)
time root -b -q extract_universes.C\(\"piminus_ON_Pb_AT_5GeV_f6n200.hist.root\",\"./extracted\"\) > extract.log 2>&1 &
- now , with directories created, for professor create multiverse parameter files